Examining the shock response of iron using molecular-spin dynamics

For magnetic materials like iron, where structural deformations are coupled to magnetic properties, resolving the underlying magnetization dynamics within the spin subsystem is critical for capturing the correct material response. Herein, utilizing a large dataset of high temperature/pressure ab-initio calculations, we introduce a novel data-driven framework for building molecular-spin dynamics machine learned potentials that can capture both phononic and magnonic excitations. We show that our molecular-spin model for iron can accurately reproduce, elastic properties, bcc-hcp (~15GPa) / bcc-fcc (~26 GPa) transition pressures, and experimental ramp compression curve data. We further test our framework by carrying out large scale molecular-spin dynamics calculations that examine the shock response of iron for piston speeds in the range of 0.1-3.5 kmps. Doing this we characterize both the phase transformation and shock-induced melting pressures. Additionally, we examine how the shock-hugoniot curves vary for different preheat temperatures. The impact of longitudinal spin fluctuations on the shock-induced phase transformation pressures is also examined.